VIEWPOINT
The expanding size of networks, enhancements in capacity, and the need for decarbonization present significant challenges for network O&M. One of the biggest challenges is the management of vast amounts of information. The emergence of Generative AI (GenAI) offers solutions to these problems. GenAI has been used to develop Autonomous Networks Agent (AN Agent) to enhance the human ability to manage large-scale complex operations in the world’s largest telecom network, China Mobile.
China Mobile’s network O&M is both costly and labor-intensive. For example, one branch of China Mobile experiences over 2,500 alarms each month, with an average ticket handling duration of over 7 hours. Additionally, the annual core network evaluation takes an entire month to complete. To address these challenges, China Mobile and Huawei have collaborated on the development of telecom large language models and autonomous networks.
Compared to traditional network fault management solutions, China Mobile Guangzhou’s new solution has three distinct core capabilities:
- Solution upgrade from alarm management to fault management
This solution streamlines the resource and service modules and uses machine learning algorithms to reduce noise in alarm data. It then associates and clusters alarms, significantly reducing the number of alarms and tickets. Additionally, the solution uses In-situ Flow Information Telemetry (IFIT) to perform association analysis on the impact of poor-QoS base stations on services, which helps determine troubleshooting priorities.
- O&M mode upgrade driven by scenario-specific foundation models
This new solution covers the entire SPN O&M process, from fault identification to analysis and rectification. It has a strong semantic understanding and can be used as an intelligent chatbot for dialog-based interaction and flexible search based on fault scenarios. This significantly reduces the need for complex manual configuration, simplifies troubleshooting procedures, and assists service decision-making. This solution increases the fault diagnosis automation rate from 60% to 80%.
- Fault closure process improved through small- and medium-screen linkage
The solution builds small- and medium-screen linkage that allows the mobile app (the small-screen assistant) to receive real-time information about network resources and faults through intent-oriented APIs. This eliminates repetitive network management tasks and ensures quick responses from both network operations center (NOC) engineers and field maintenance engineers by freeing them from frequent communication through chat apps and phone calls. It also makes the troubleshooting process visible and manageable, significantly shortening the average fault response time by 83%.
We are conducting region-specific tests and have received positive feedback on the effectiveness of the AN Agent solution for telecom digital transformation. This feedback indicates that we can provide an enhanced network experience for a significant number of users. This success is particularly promising for our city network support efforts, as it allows us to safeguard the execution of large-scale exhibitions and concerts effectively.
